create-llm CLI
The CLI scaffolds language-model training projects with tokenization, data preparation, checkpoints and evaluation.
- 373 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The generator creates files and installs dependencies; subsequent training consumes machine resources and may use configured external integrations.
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/create-llmDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The generator creates files and installs dependencies; subsequent training consumes machine resources and may use configured external integrations.
This is third-party code. Review the repository files before installing.
What it does
create-llm scaffolds a language-model training project. Users choose a template size and tokenizer, then receive data preparation, training, evaluation and text-generation code. A shared configuration file controls the model and training settings. The project includes checkpoints and TensorBoard monitoring. Users add text, train the tokenizer and launch training separately; generating folders does not produce a trained model.
Who it is for. Educators and ML practitioners setting up a small training experiment on their own text data.
Good fit when
- For a lesson, you want to show the complete path from text to generation.
- With your own dataset, you need to test tokenization and training.
- Before larger training, you prepare a small CPU experiment.
- In a teaching project, you need ready evaluation and training-resume commands.
Not a fit when
- You expect a frontier-quality model immediately after installation.
- The selected template exceeds available data or memory.
- The training texts’ license does not permit model training.
Example request
Create a small educational create-llm project and show data preparation steps before training.Limitations
The CLI needs Node.js 18 or later and npm 8 or later; check the environment before scaffolding. Training separately requires Python 3.8 or later and PyTorch 2.0 or later. Larger templates need substantially more memory and data, so validate a small configuration on your machine first. Quality depends on the corpus and training setup; the author’s template sizes and timing estimates do not guarantee useful results on your text.
How to disable. Stop training and monitoring, then preserve needed checkpoints. Removing the CLI does not remove generated projects or dependencies.
FAQ
Where does training text go?
Place it in `data/raw/` of the generated project; data preparation runs separately.
How do I open the interactive wizard?
Run `npx create-llm` without a project name.
Can I skip dependency installation?
Yes. The generator supports `--skip-install`.
Where do training settings live?
In `llm.config.js`, alongside model configuration.
How do I monitor training?
Use TensorBoard with `logs/tensorboard` and inspect metrics alongside validation results.
Discussion
No questions or reviews yet.
Related
CLI toolsA CLI and MCP server that use set rules to pack only the project files a task needs for the model
A skill that adds the local Laya model to a project: it classifies text and returns probabilities with no LLM call.
A coding agent for small, cheap LLMs: splits a goal into a task tree with contracts and verifies each leaf with builds and tests.
Commands /implement, /design and /build, specialized agents and skills for Claude Code in a TypeScript project on Biome and Vitest
LLM Sandbox MCP: an agent runs code in seven languages in a Docker, Podman or Kubernetes container and gets plots.
Linghun is a terminal coding agent built on evidence: it reads code before answering, verifies edits with tests and checkpoints git before changes
README in short
Templates cover several sizes and tokenizers. Separate scripts support interactive chat and later model deployment.